ARTICLE pubs.acs.org/est
Formation and Structural Characterization of Thioantimony Species and Their Natural Occurrence in Geothermal Waters Britta Planer-Friedrich*,† and Andreas C. Scheinost‡ † ‡
Environmental Geochemistry, University of Bayreuth, Universitaetsstrasse 30, 95440 Bayreuth, Germany The Rossendorf Beamline (BM20), ESRF, 6 Rue Jules Horowitz, Grenoble 38043, France, and Institute of Radiochemistry, HZDR, Bautzener Landstrasse 400, 01314 Dresden, Germany
bS Supporting Information ABSTRACT: Previously postulated from laboratory studies, the occurrence of antimonysulfur species in geothermal waters could now be proven using anion-exchange chromatographyinductively coupled plasmamass spectrometry. The two thioantimony species detected by AEC-ICP-MS in oxic synthetic antimonite-sulfide solutions were assigned to tri- and tetrathioantimonate based on their S/Sb ratios and structural characterization by X-ray absorption spectroscopy (XAS). XAS confirmed that the initial species formed under anoxic conditions from antimonite at a 10-fold sulfide excess is trithioantimonite. Trithioantimonite rapidly transforms to tetrathioantimonate in the presence of oxygen or to antimonite at excess OH versus SH concentrations, and escapes chromatographic detection. In natural geothermal waters, up to 30% trithioantimonate and 9% tetrathioantimonate were detected. Their occurrence increased at increasingly alkaline pH and with increasing sulfide and decreasing oxygen concentrations. Considering the large sulfide excess (100 to 10 000-fold) the proportion of thioantimonates formed under natural conditions is lower than expected from synthetic solutions. Together with the observed high thioarsenate concentrations (>80% of total arsenic), this indicates that in direct competition with arsenic for a limited source of sulfide, thioantimonates form less spontaneously than thioarsenates. Interactions of arsenic and antimony with sulfur can therefore be decisive for similarities or differences in their environmental behavior.
’ INTRODUCTION Antimony (Sb) does not exceed concentrations of 1 μg/L in most aquatic environments.1 However, it is commonly enriched in hydrothermal deposits as stibnite (Sb2S3), tetrahedrite ((Cu, Fe)12Sb4S13), and berthierite (FeSb2S4). Dissolution of these minerals can lead to substantially higher antimony concentrations in geothermal waters (e.g., up to 2.5 mg/L at El Tatio Geyser Field, Chile2). In numerous laboratory studies, antimony speciation upon stibnite dissolution has been measured or modeled under a wide range of pH and temperature conditions. The importance of aqueous antimony sulfide complexes, so-called thioantimony species, for dissolution and transport of antimony in geothermal solutions has often been pointed out and is assumed to be linked to the mobility of other elements of economic (Au, Ag, Cu, Pt, Pd) or toxicological (As) interest. The prevailing view, up to the late 1990s, was that only Sb(III) thiocomplexes are important for geothermal antimony transport, and that predominantly multimeric species form upon dissolution of stibnite. The dimeric species HSb2S4 and Sb2S42 were considered the stoichiometrically most probable and thermodynamically most stable thioantimonite complexes at lower temperatures,36 while at temperatures >120 °C the hydroxothioantimonite complex Sb2S2(OH)20 should form.3,7 r 2011 American Chemical Society
Wood6 cautioned that in geothermal fluids, antimony and sulfide concentrations are significantly lower than in most solubility studies and that thus monomeric (SbS33; or—based on ab initio calculations7—SbS2(SH)2 and SbS(SH)2) rather than polymeric species may dominate. More recent studies suggest that due to the low standard reduction potential of 0.6 V,8 thioantimonites quickly oxidize to thioantimonates. The formation of sulfidic Sb(V) complexes could thus explain previously puzzling findings of Sb(V) occurrence in anoxic environments.9,10 Two oxidation mechanisms have been suggested based on elemental sulfur in polysulfides or protons as oxidizing agents to finally yield—besides thioantimonates—shorter chain polysulfides or H2 gas, respectively: SbðIIIÞ þ Sn 2 þ Hþ f SbðVÞ þ HS þ Sn1 2 11 ð1Þ SbS3 3 þ HS þ Hþ f SbS4 3 þ H2 ðgÞ12
ð2Þ
Received: March 25, 2011 Accepted: July 12, 2011 Revised: July 1, 2011 Published: July 12, 2011 6855
dx.doi.org/10.1021/es201003k | Environ. Sci. Technol. 2011, 45, 6855–6863
Environmental Science & Technology Based on the latter equation it was deduced12 that the seemingly contradictory observations of thioantimonites in solubility studies, versus thioantimonates in spectroscopic studies, are inherent to the analytical design of each method. Whereas solubility studies are usually carried out at low sulfide concentrations to avoid activity coefficient uncertainties, spectroscopic studies use high sulfide concentrations to maximize the spectral signals, which pushes the equilibrium in eq 2 toward thioantimonate formation. A discrepancy continues to exist about the dominance of monomeric versus polymeric thioantimonates. Two studies published in 2000 found tetrathioantimonate (reported as Sb(HS)4+ 13 or SbS43 12) was the dominant species below 150 °C. Polymeric species were only reported at temperatures exceeding 250 °C,13 or in a few solutions but without correlation to total Sb concentrations or pH.12 A recent study11 reports the dimeric HSb(III)Sb(V)S5 and Sb(V)2S62 species to be the dominant thioantimony species in solutions equilibrated with elemental sulfur. Considering the number of theoretical studies suggesting that the formation of thioantimonites and thioantimonates will have a substantial effect on geothermal antimony transport, it is all the more surprising that, to the best of our knowledge, there are no published data on the occurrence and fate of thioantimony species in the environment. Once more, the knowledge on antimony lags behind its “notorious group 15 cohort”11 arsenic as the occurrence of thioarsenates in geothermal waters has been studied intensively in the last five years,14 along with their abiotic and microbial-catalyzed transformation15 and their stability compared to thioarsenites.1620 One major problem for the determination of thioantimonates in aquatic environments is that the spectroscopic techniques currently applied for synthetic solutions, XAS and RAMAN, are not suitable for studying antimony speciation at the typically low natural concentrations. For the present study, we investigated the applicability of anion-exchange chromatography for thioantimony speciation in synthetic solutions and geothermal waters and successfully applied it to determine the natural occurrence of thioantimonates at a wide range of pH, temperature, sulfide, oxygen, and arsenic concentrations.
’ MATERIALS AND METHODS Experiments with Synthetic Antimony Solutions. To evaluate the chromatographic separation and mass-spectrometric detection method, several synthetic solutions were prepared of antimonate (KSb(OH)6, Fluka, dissolved by 30 min ultrasonication), antimonite (C4H2KO6Sb 3 1.5H2O, Acros Organics), sulfide (Na2S 3 9H2O, Sigma-Aldrich), and thiosulfate (Na2S2O3 3 5H2O, Applichem). SodiumEDTA (C10H14N2Na2O8 32H2O, Gr€ussing) was added to selected antimonite solutions to test its efficiency as a complexing agent to enhance antimonite elution. To investigate the conditions required for the formation of antimonysulfur species, sulfide was added to 0.1 mM antimonite solutions to yield final S/Sb ratios of 1, 2, 5, 10, or 20 either inside a glovebox containing 95%N2/5%H2 atmosphere (“anoxic samples”) or outside under normal atmosphere (“oxic samples”). For a S/Sb ratio of 10, one pH series and two dilution series were prepared. In the pH series, a 0.1 mM antimonite solution was adjusted to a nominal pH of 3, 4, 6, 7, 9, and 11. The actual sample pH was measured with a HACH pH meter HQ 40d. In the dilution series, final Sb concentrations were 0.001, 0.005, 0.01, 0.05, and 0.1 mM. The first dilution series was prepared under anoxic conditions using deionized
ARTICLE
water as diluting agent, which also changed the final pH of the diluted solution. A second and a third dilution series were prepared in 0.5 mM NaOH to maintain a pH of approximately 11 under oxic and anoxic conditions, respectively. For structural characterization by XAS, experimental concentrations were scaled up to 10 mM antimonite, and 20, 100, and 200 mM sulfide were added. The anoxic samples were pipetted immediately after mixing into PE sample holders for XAS analysis and covered with Kapton tape inside a glovebox. Together with oxic samples prepared under ambient air conditions, the anoxic samples were flash-frozen in liquid nitrogen outside the glovebox immediately after preparation, stored in the freezer and transported on dry ice from the preparation laboratory in Bayreuth to the Rossendorf Beamline (BM20) at the European Synchrotron Radiation Facility (ESRF) in Grenoble. AEC-ICP-MS Analyses. Antimony speciation was conducted as described previously for thioarsenates14 by anion-exchange chromatography (AEC) with an AG16/AS16 IonPac column using an alkaline gradient (20100 mM NaOH) at a flow rate of 1.2 mL/min. The routine setup was an ICS-3000 SP (Dionex) installed outside the glovebox, coupled to an inductively coupled plasmaquadrupole mass spectrometer (ICP-MS; X-Series2, Thermo Scientific). Using 10% oxygen in 90% helium as ICP reaction gas, sulfur was measured as SO+ at m/z 48 and arsenic was measured as AsO+ at m/z 91. Antimony was initially measured as Sb+ at m/z 121 and 123 and as SbO+ at m/z 137 and 139; later routinely only at m/z 121. Thioantimonates were quantified for their sulfur content based on sulfate ((NH4)2SO4, Fluka) calibration curves and for their antimony content based on antimonite calibration curves. Antimonate was calibrated separately. For selected anoxic experiments, an HPLC gradient pump (System 525, BioTek Instruments) was set up inside the glovebox coupled to the ICP-MS (outside the glovebox) to exclude any traces of oxygen during chromatographic separation. Sample injection and gradient changes were done manually. XAS Analyses. Antimony K-edge (30491 eV) XAFS spectra were collected at the Rossendorf Beamline at the European Synchrotron Radiation Facility (ESRF), using a Si(111) doublecrystal monochromator. Third-order harmonics were suppressed by using two 1200-mm-long mirrors with Pt coating. Spectra were collected either in transmission mode using gas-filled ionization chambers or in fluorescence mode using a high-purity 13element Ge solid-state detector (Canberra) with a fast digital spectrometer (XIA). The samples were measured in a closed-cycle He cryostat (CryoVac) at 15 K to maintain the Sb oxidation state during the measurement (up to 12 h) and to eliminate thermal contributions to the DebyeWaller term, thereby greatly improving the detection of structural features beyond the first coordination shell.21 The photon energy was calibrated against the absorption edge of an Sb0 foil measured in serial arrangement behind the sample. Averaging of several spectra, energy correction, and fluorescence deadtime correction were performed with SixPack.22 Normalization, conversion into k-space, spline background removal with autospline, Fourier transformation (Bessel windows with smoothing factor 3), and shell fit were performed using standard procedures in WinXAS.23 Theoretical phase shift and amplitude functions were calculated with FEFF8.2.24 Collection and Analyses of Geothermal Water Samples. Natural samples were collected in July and August 2009 at 19 sites in Yellowstone National Park, 7 sites in Steamboat Springs, Colorado, and 2 sites in Steamboat Springs, Nevada. In Yellowstone National Park, drainage profiles of numerous 6856
dx.doi.org/10.1021/es201003k |Environ. Sci. Technol. 2011, 45, 6855–6863
Environmental Science & Technology sampling points with increasing distance from the source of the hot spring were sampled at six sites: in the Lower Geyser Basin at Boulder Geyser (8 sampling points), Conch Spring (10), Ojo Caliente (12), and Flat Cone (7); in the Gibbon Geyser Basin at GGSNN011 (13); and in the Joseph's Coat Basin at Scorodite Spring (8). In total, 80 geothermal water samples were collected covering a wide range of pH (29.4), temperature (2294 °C), redox (65 to 240 mV), and oxygen (897% saturation) conditions, total mineralization (4009050 μS/cm), total sulfide (mean 1.4 mg/L, maximum 16.3 mg/L), arsenic (mean 2.7 mg/L, maximum 15 mg/L), and antimony (mean 50 μg/L, maximum 970 μg/L) concentrations (Table SI1). Conductivity, pH, and redox potential were measured using a TetraCon 325 conductivity cell, MultiLine P4 with a SenTix 97/T pH electrode, and a Pt 4805/S7 probe (WTW, Germany). Dissolved oxygen was measured with a HACH HQ20 luminescent dissolved oxygen sensor. Sulfide was determined photometrically according to the HACH methylene blue (no. 8131) method. Samples for As and Sb speciation were filtered with a 0.2-μm cellulose-acetate filter (Membrex), flash-frozen on dry ice on-site, and kept frozen until analysis. Prior to AEC-ICP-MS analysis samples were thawed in a glovebox and analyzed within a maximum of 12 h outside the glovebox. To increase the long-term plasma stability during analysis of large batches of natural samples, an anion-self-regenerating suppressor (ASRS 4 mm) was used postcolumn. Thioantimonate quantification was based on the antimonate calibration curve. Total arsenic and antimony concentrations were determined by ICP-MS analysis in samples diluted 1:10 with deionized water. Antimonite concentrations were calculated as the difference between total antimony concentrations and the sum of concentrations of antimonate and thioantimonates concentrations. We acknowledge that, due to the lack of an independent antimony speciation balance, any further antimony species—if present in solution and not distinguishable by our chromatographic separation—will remain unrecognized and be summarized falsely as antimonite. As the focus here was the first identification of thioantimonates in natural waters we prioritized general instrument long-term stability over accurate speciation balance.
’ RESULTS AND DISCUSSION Evaluation of the AEC-ICP-MS Method for Antimony Speciation. Using the AEC-ICP-MS method previously optimized
for thioarsenates14 we were able to successfully determine antimonate, antimonite, and thioantimonates. Antimonite is generally difficult to determine by anion chromatography. From pH 1.5 to 11.9 it predominates as uncharged Sb(OH)3 complex and should thus be eluted in the solvent front. However, irreversible retention on the column has also been reported. For example using a SbCl3 standard may cause antimony oxychloride or hydroxide precipitation and using a Sb(III) tartrate standard may cause strong binding of the dinegative ion.25 Successful chromatographic separation requires prior complexation, e.g., with phthalic, malonic, tartaric acid26 or EDTA,25 to form a negatively charged antimonite ion and the addition of a strong competing anion to the eluent for desorption.25 In our method, because of the high eluent pH, with a gradient from 20 to 100 mM NaOH, antimonite predominates as negatively charged Sb(OH)4 ion (70 to 92%, respectively, based on modeling using database version thermo. com.V8.R6.23027 ). As previously observed,25 antimonate (retention time (RT) = 140160 s) is eluted before antimonite
ARTICLE
(RT = 240260 s) (Figure 1). Compared to antimonate, an antimonite solution of equal molality yielded peak area signal intensities of 93 ( 10% (n = 25, Table SI 2a). The addition of 520 mM EDTA did not significantly increase the elution efficiency (Table SI 2b) and was not pursued. For arsenic speciation,14 an anion-self-regenerating suppressor was used postcolumn to electrochemically neutralize the hydroxide eluent and remove the counterion (Na+) by exchange with H+. This reduces the salt load introduced into the plasma resulting in improved detection limits and better long-term stability. When using the suppressor for antimony speciation, antimonite was lost (Table SI 2c), probably because lowering the pH led to the transformation of Sb(OH)4 to Sb(OH)3, which is subject to exchange or precipitation. An analytically complete antimony balance including antimonite could only be obtained without suppressor. The downside is a slightly lower absolute signal intensity for antimonate in synthetic solutions (Table SI 2c) and an about 50% lower absolute signal intensity for antimonate and thioantimonates in synthetic SbS solutions (Table SI 2d) compared to analyses with the suppressor. Besides antimonate and antimonite, two antimony species with retention times of 11201150 s and 12101240 s, respectively, could be separated both in synthetic antimonite-sulfide solutions and in natural geothermal waters (Figure 1). Simultaneous peaks at the respective retention times in the sulfur track indicate that these species are antimonysulfur species. The S/Sb ratios calculated for different synthetic solutions were 3.08 ( 0.28 (n = 19) and 4.05 ( 0.32 (n = 18) (Table SI 2e), which match the ratios expected for tri- and tetrathioantimonate. Structural characterization by X-ray absorption spectroscopy and preliminary information about the conditions for their formation are described below. In natural samples the calculation of S/Sb ratios was hampered by interferences of tetrathioarsenate-sulfur (RT = 11101140 s) and potentially other sulfur species as reported previously14 or by very low thioantimonate concentrations, which were still detectable in the antimony track but not in the less sensitive sulfur track. To avoid interferences on m/z 32 from 16O2+ for the determination of sulfur 32S+ with a low-resolution quadrupole ICP-MS, oxygen was added as a reactive gas and sulfur was measured at m/z 48 as 32S16O+. Also antimony reacts with oxygen to form SbO+, unfortunately however, not quantitatively. Table SI 3 shows that, in standard mode, 56.5 ( 0.04% of the total signal intensity was measured on m/z 121, which is close to the natural relative isotope abundance for 121Sb/123Sb of 57.4/42.6.28 Applying the reactive oxygen mode, formation of 4.5 ( 0.12% 121 16 + Sb O at m/z 137 and 3.4 ( 0.02% 139Sb16O+ at m/z 139 was observed. The highest signal intensity remained at m/z 121 with 52.0 ( 0.05%. This relative loss in signal intensity was more than compensated for by an overall increased sensitivity. The absolute signal intensity for 121Sb was 2.4 times higher in reactive oxygen mode than in standard mode. Structural Characterization of Thioantimony Species by XAS. To characterize the two antimonysulfur species observed by AEC-ICP-MS, synthetic antimonite solutions were prepared with a 2-, 10-, and 20-fold excess of sulfide and analyzed by XAS. The respective antimony K-edge X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine-structure (EXAFS) spectra are shown in Figure 2a. The three spectra grouped at the top are from solutions prepared under normal oxic atmosphere (samples S57) and the three spectra grouped in the middle are from solutions prepared in the anoxic glovebox 6857
dx.doi.org/10.1021/es201003k |Environ. Sci. Technol. 2011, 45, 6855–6863
Environmental Science & Technology
ARTICLE
Figure 1. Example chromatogram from alkaline exchange chromatography showing the Sb, As, and S tracks with resolution of 4 Sb, 6 As, and 8 identified S peaks (1 = antimonate, 2 = antimonite, 3 = trithioantimonate, 4 = tetrathioantimonate, 5 = arsenite, 6 = arsenate, 7 = monothioarsenate, 8 = dithioarsenate, 9 = trithioarsenate, 10 = tetrathioarsenate, 11 = sulfide, 12 = sulfate, 13 = thiosulfate, 1416 = mono-, di-, trithioarsenate, 17 = tetrathioarsenate and trithioantimonate, 18 = tetrathioantimonate).
(samples S24). Within both groups, S/Sb ratios increase from bottom to top. The spectrum of the solid Sb2S3 reference is shown at the bottom. The k3-weighted EXAFS spectra were analyzed by iterative transformation factor analysis (ITFA).21,29 The set of 7 spectra required four components to be reproduced within the noise level (not shown). Figure 2b shows these four factors in k and R space, and the corresponding factor loadings for the samples. Factor 1 represents a first coordination shell, which is most purely expressed in the oxic samples with S/Sb ratios of 10 and 20 (S6, S7). This shell is fitted by about four SbS paths (CN 4.24.3) of 2.332.34 Å long (Table SI4) in line with a pentavalent SbS species.12,13 Factor 2 represents a slightly larger coordination sphere, which is most purely expressed in the anoxic samples with S/Sb ratios of 10 and 20 (S3, S4). This bond length, together with a coordination number of three, has previously been interpreted as trivalent SbS aquo species.12 However, our SbS coordination number varies between 3.4 and 3.7, and both the SbS distances and the edge energies lie in between those of factor 1 (Sb(V)-S) and stibnite (Sb(III)-S), hence could also be interpreted as a mixed-valence Sb(III)Sb(V)-species. Such a species, HSb2S5, has been identified in a previous study.11 However, our data do not reveal any SbSb interaction as expected for this species, although dynamic disorder— which could potentially wash out second shell interactions due to destructive interference—was suppressed by collecting the EXAFS spectra at 15 K. Given the high uncertainty of EXAFS coordination numbers (plus/minus 25%), and the influence of coordination geometry on edge energies, we are in favor of interpreting factor 2 as trithioantimonite, although we cannot completely exclude the interpretation as a mixed-valence species. Factor 3 represents an even longer coordination distance, and in addition a significant contribution of backscattering atoms beyond the first coordination sphere. This feature is expressed only in the Sb2S3 reference with S backscattering contributions at 2.51 Å and 3.09 Å, and with Sb contributions at 3.82 Å and 4.01 Å, in line with previous descriptions of the structure of stibnite.30 Although freely fitted coordination numbers are much smaller than expected, in line with similar observations for Sb oxide minerals,21 their Fourier transform magnitude shows backscattering contributions up to 6 Å, indicative of a molecular
structure extending across at least this distance. Factor 4 is dominated by a FT peak at the very short distance of about 1.5 Å (uncorrected for phase shift) and is most purely expressed in the oxic sample with an S/Sb ratio of 2 (S5). The shortest shell of these samples is fitted by about one O atom at 1.96 Å, a typical distance for oxygen coordination to both Sb(III) and Sb(V).21 The XAS data clearly show that, under anoxic conditions, the trivalent oxidation state of the original antimonite is preserved and excess sulfide leads to formation of trithioantimonite. Due to the absence of longer-range backscattering contributions, there is no evidence for the formation of polymeric species such as the dimeric species HSb(V)Sb(III)S5 or Sb(V)2S62, which have been observed in a previous study on stibnite solubility in the presence of zerovalent sulfur.11 The result is consistent with an earlier conclusion6 that the formation of polymeric species is unlikely in solutions with antimony concentrations below 0.1 M. At low sulfide concentrations, there is still a contribution from an SbO coordination. Whether this spectrum constitutes a mixed SbSO species, or the presence of two species, one with only O and the other with only S coordination cannot be determined from the EXAFS data. Under oxic conditions, tetrathioantimonate forms at excess sulfide. As for the anoxic systems, O is present in addition to S in the Sb first coordination sphere at the lowest S/Sb ratio. This sample has conserved the initial Sb(III) oxidation state as indicated by the SbS distance of 2.40 Å, and also shows a SbSb path of 5.64 Å. Both in oxic and anoxic samples at S/Sb ratios g10, only one thioantimony species per sample was detectable by EXAFS. Comparability of AEC-ICP-MS and XAS Results. The results of analyzing antimonite solutions with a 1-, 2-, 5-, 10-, and 20-fold excess of sulfide by AEC-ICP-MS are shown in Figure 3 and Table SI5. The total concentrations are 100 times lower than those in the solutions used for XAS characterization. Under oxic conditions (Figure 3a), at S/Sb ratios of 10 and 20, the antimony sulfur species with RT 12101240 s accounts for more than 90% of the Sb in solution. This species has provisionally (based on S/Sb ratios of 4.05 ( 0.32, see above) been assigned as tetrathioantimonate. The XAS data (samples S3, S4; Figure 2, Table SI4) confirm this identification. In the same solutions, we detected about 89% of the antimonysulfur species with RT 6858
dx.doi.org/10.1021/es201003k |Environ. Sci. Technol. 2011, 45, 6855–6863
Environmental Science & Technology
ARTICLE
Figure 2. XAS spectra for antimony speciation in synthetic anoxic (samples S2S4) and oxic (samples S5S7) Sb(III)-S- solutions. (a) SbK X-ray absorption spectra of SbS samples and references. Left: XANES; center: EXAFS; right: corresponding Fourier transform. (b) Left: The ITFA-derived first four Eigen vectors of the EXAFS spectra shown in (a) in R-space and in k-space (insert). Right: Varimax factor loading of factors 1 to 4.
11201150 s, which has provisionally been assigned as trithioantimonate (S/Sb ratio 3.08 ( 0.28). No evidence for the occurrence of trithioantimonate could be derived, however, from the XAS data. This suggests that the trithioantimonate percentage is too low in relation to the dominant tetrathioantimonate to be detected by XAS (
